kananvyas/boxdetection

A Box detection algorithm for any image containing boxes.

A Box detection algorithm for tabularized data.

• This code is used to extract data which is in tabular format using image processing techniques.
• When you are working with Optical character recognition(OCR) or any data or object recognition problem, the first thing to do is preprocessing.
• Here preprocessing means to extract the box where our data is located. After extracting the boxes, any OCR algorithm can be performed on those crops for recognition.

• The problem arises when you have to detect objects which are located in any tables/boxes or in row-column format. If the image is like this then you have to detect boxes and extract them one by one. Now it should be done accurately for all images.

This algorithm helps to detect every boxes accurately and save it in a /Output/ folder. The code is shown in src/box_detection.py

USAGE:

• Run python src/box_detection.py
• You can see the output crops on /Output folder

You can also read the medium article for understanding about the algorithm: https://medium.com/@kananvyas/a-box-detection-algorithm-for-any-image-containing-boxes-756c15d7ed26

boxdetection's Issues

ModuleNotFoundError: No module named 'cv2'

May seem obvious for somewhat seasoned developers.
But cv2 isn't installed by default.

A requirements.txt with the dependencies would already help a lot in setting up a virtual environment & getting started quicker.

OCR Detection using Box Co-ordinates

Can you please tellme how to extract text using co-ordinates of the rectangular box?
I tried varying the kernel length for the same.
I run the same code for different pic image but I cannot see any image being created in Cropped folder

i am getting img_bin as null type img_bin = 255 - img_bin # Invert the image TypeError: unsupported operand type(s) for -: 'int' and 'NoneType'

extracting boxes from tambola ticket failed to separate 8th and 9th column

Please see attached image and output is

My aim is to extract each box then read a number from it if it is not empty .
Can you suggest ?

Thanks.

What could be possible reason for ValueError

Find contours for image, which will detect all the boxes

im2, contours, hierarchy = cv2.findContours(img_final_bin, cv2.RETR_TREE,cv2.CHAIN_APPROX_SIMPLE)

ValueError: not enough values to unpack (expected 3, got 2)

How to handle the situation where the line is recognized as di-continuous

Hi, I encounter a problem where a line is recognized as 2 separate lines, so it gives trouble for the contour recognition. Do you have some suggestions on how can we handle this?

Can you please provide installed libraries version..!

Better if you add requirements file to resolve dependency issues.

Extract a boxe from an image

hi, i need to extract the two red boxe and i would use your code to do it but when i execute it i have a bad result. Down here my image

Even if i extracted one only big red boxe instead of two smaller, it's ok still.
Can you please help me?

Missing License

Hi, thank you for the excellent medium article and this repository.
Please add a license to your project.

https://help.github.com/en/github/building-a-strong-community/adding-a-license-to-a-repository

Thanks

not getting the exact rectangular box i need.

its working for me for some images and not working for all

here in the image it may have one circle shape and two or three rectangle boxes and in that three i need the bigger one.

if there is only one rectangle its getting but when there is two or more it giving another rectangle of different size.